Publication details
Algorithms v/s Data: Thinking the Machinic Reconfigurations of Governmentality
| Authors | |
|---|---|
| Year of publication | 2026 |
| Type | Appeared in Conference without Proceedings |
| MU Faculty or unit | |
| Citation | |
| Description | Since the 2010s—with the dramatic resurgence of neural networks in machine learning, driven by their use for computer vision—critical research on AI has begun to redirect its focus from algorithms to data and data practices. Several authors point, in this regard, to a mutation that is not merely of scope (the data-hungry nature of neural networks) but represents a “paradigm shift” in machine learning (Mackenzie, 2017; Malevé, 2021). This shift has significant implications for the exploitation of labor (the expansion of gig tech platforms for data labeling and their distinctive mechanisms of profiting from and amplifying inequities), but also (jointly) for the reconfiguration of a certain “orchestration of reality” (Gourinova, 2025). My presentation will aim to identify some key aspects of this shift through the lens of the Foucauldian concepts of biopolitics (Foucault, 2003; Foucault, 2016; Foucault, 2007; Foucault, 2008) and anatomo-politics (Foucault, 1977; Foucault, 2012). The aim will be to bring to light the reconfigurations of governmentality between algorithms and data, focusing on the implications of the changes related to modeling in machine learning. In other words, the move from modeling as an engineering design process to modeling as driven by the data themselves, that is, by function-extracting machines traversing enormous curated datasets. The case of emotion AI in its relation to computer vision will serve as the domain of reference. I will analyze the data practices that have shaped the assemblage of the analytical repository of the tech company Affectiva (McDuff, Kaliouby, & Picard, 2011; Kaliouby, 2019)—the first, and to date, the largest emotion data repository. My aim is to propose, through the framework of governmentality, a critique that can operate at the articulation of the genealogies of algorithms and an analysis of the current reconfiguration of data practices. |
| Related projects: |